hnf1b Genes in Zebrafish Hindbrain Development

Abstract

The Hnf1b transcription factor acts during formation of rhombomeres (r) 5 and 6 in the hindbrain. To determine if hnf1b is absolutely required in r5/r6, we examined the hnf1b^hi2169 and hnf1b^hi1843 retroviral insertion alleles. Hnf1b^hi2169 shows highly variable residual expression of several genes in r5/r6, but this is not due to full-length hnf1b transcripts persisting in hnf1b^hi2169 embryos, nor to hnf1bl, a novel hnf1 family member expressed in r5 that we identified. Instead, we find evidence for a virus-hnf1b fusion transcript in hnf1b^hi2169 embryos and demonstrate that morpholino-mediated knockdown of this transcript leads to near-undetectable r5 gene expression. The hnf1b^hi1843 allele has a more severe phenotype with near-undetectable expression of r5/r6 genes. We next examined if hoxb1b, which functions upstream of hnf1b in r5/r6 formation, can induce expression of r5/r6 genes in hnf1b mutants. We find that microinjected hoxb1b mRNA induces ectopic gene expression anterior to the hindbrain in hnf1b^hi2169 and hnf1b^hi1843 embryos, but cannot restore gene expression in r5/r6 of the mutants. We conclude that hnf1b^hi2169 is hypomorphic to hnf1b^hi1843 and that, while hnf1b is required for r5/r6 gene expression in the hindbrain, r5/r6 gene expression can be experimentally induced independently of hnf1b anterior to the hindbrain.

Introduction

The vertebrate hindbrain is transiently subdivided into rhombomeres during embryonic development. The segmented nature of the embryonic hindbrain ensures proper spatial positioning of differentiating neurons (e.g., reticulospinal interneurons and motornuclei of the cranial nerves) and also patterns neural crest cells that migrate from the hindbrain and contribute to numerous structures (e.g., peripheral nervous system and craniofacial structures). Ongoing efforts to understand hindbrain patterning have identified a number of genes required for rhombomere formation.¹ In particular, several genes have been implicated in the formation of rhombomere (r) 5 and/or r6, including the homeodomain transcription factor variable hepatocyte nuclear factor 1/transcription factor 2/hnf1 homeohomeobox b (vhnf1/tcf2/hnf1b), the bZip transcription factor mafB/kreisler/valentino (mafB/Kr/val), the zinc finger transcription factor krüppel box 20/early growth response 2 (krox20/egr2), and homeodomain transcription factors from hox paralog group 3 (PG3 hox).

Previous work has suggested a pathway wherein hnf1b is required for activation of val, which in turn is required for activation of krox20 and PG3 hox genes. Accordingly, hnf1b expression precedes val expression, val expression is disrupted in r5-r6 of hnf1b mutants^2–4 and the val promoter contains functional hnf1b binding sites.⁵ Further, essential val (mafB) binding sites are present in the hoxa3 and hoxb3 promoter region,^6–8 and expression of hoxa3 and hoxb3 is absent from r5-r6 in kreisler mutant mice and valentino mutant zebrafish.^9–12 Lastly, krox20 expression is disrupted in r5 of both hnf1b and val mutant zebrafish,^2–4,13 as well as in kreisler mutant mice.⁹ However, more recent work has brought this proposed simple linear regulatory cascade into question. For instance, it appears that hnf1b may directly activate krox20 expression without requiring val.¹⁴ In addition, mafB proteins may feed back to maintain hnf1b expression,¹⁵ suggesting that these genes may constitute a regulatory network. Further, several aspects of the published hnf1b mutant phenotype, including val expression in r5/r6, appear to be variable,² suggesting that hnf1b may not be absolutely required for val expression or r5/r6 formation.

Here we characterize the phenotype of the hnf1b^hi2169 retroviral insertion allele in greater detail. We find highly variable residual expression of several r5/r6 genes in hnf1b^hi2169 embryos. This residual gene expression is not the result of full-length hnf1b transcripts persisting in hnf1b^hi2169 embryos, nor it is due to hnf1bl, a novel hnf1 family member expressed in r5 whose activity is indistinguishable from that of hnf1b. Instead, we find evidence for a virus-hnf1b fusion transcript. Although we lack antibodies to directly detect the presence of a protein produced from this transcript, we demonstrate that morpholino-mediated knockdown of this transcript leads to a stronger phenotype with near-undetectable gene expression in r5. Turning next to the hnf1b^hi1843 allele, we find that it has a more severe phenotype with near-undetectable expression of several r5/r6 genes. We conclude that hnf1b^hi2169 is hypomorphic to hnf1b^hi1843 and that hnf1b is required for gene expression in r5/r6. Lastly, we find that hoxb1b, which functions upstream of hnf1b in the regulation of r5/r6 formation,¹⁶ can induce ectopic expression of r5/r6 genes anterior to the hindbrain in hnf1b^hi2169 and hnf1b^hi1843 embryos, but cannot restore r5/r6 gene expression to the hindbrain of these mutants, indicating that hnf1b is required for induction of r5/r6 gene expression in the hindbrain, but not further anteriorly in the embryo.

Results

Variable gene expression in the caudal hindbrain of hnf1b^hi2169 embryos

We have noted significant variations in the expression of several genes in r5 and r6 of hnf1b^hi2169 mutant embryos. The phenotype ranges from near-complete loss of krox20 gene expression in r5 (Fig. 1E) to relatively robust expression (Fig. 1B). Similarly, hoxa3 expression, which is normally observed in r5 and r6, appears completely lost in the hindbrain of some hnf1b^hi2169 embryos (Fig. 1J), but is readily detectable in other hnf1b^hi2169 embryos (Fig. 1G, H). In a representative experiment we find that only 5% (6/120) of embryos from an hnf1b^hi2169 incross lack r5/r6 gene expression. This is significantly less than 25%, indicating that most homozygous hnf1b^hi2169 mutant embryos retain r5/r6 gene expression.

FIG. 1.

Variable gene expression in the hindbrain of hnf1b^hi2169 mutant embryos. Wild-type (A, F, K) and hnf1b^hi2169 (B–E, G–J, L–O) embryos were assayed for expression of krox20 (A–E), hoxa3 (F–J), and hoxb1a (K–O) by in situ hybridization. All panels are dorsal views of flat-mounted hindbrains with anterior to the top. Rhombomere numbering is indicated in panels (A), (F), and (K). White bars in panels (N) and (O) indicate regions of reduced hoxb1a expression.

It has been reported that hoxb1a expression expands from r4 into the caudal hindbrain of hnf1b^hi2169 embryos.⁴ We observe variations also in hoxb1a expression among hnf1b^hi2169 embryos. In particular, while most hnf1b^hi2169 embryos show uniform hoxb1a expression throughout the caudal hindbrain (Fig. 1L, M), some embryos display reduced hoxb1a expression in the r5/r6 region (Fig. 1N, O). We hypothesize that embryos with reduced hoxb1a expression in r5/r6 correspond to the ones with higher expression of r5/r6 genes, since several r5/r6 genes are reported to repress hoxb1 expression.^17–19

We conclude that hnf1b^hi2169 embryos retain variable r5/r6 gene expression. These findings do not distinguish whether residual r5/r6 gene expression in hnf1b^hi2169 embryos is due to this allele retaining some hnf1b activity, or whether hnf1b is not absolutely required for r5/r6 gene expression.

Hnf1bl expression in r5 does not compensate for hnf1b activity in hnf1b^hi2169 mutants

The hnf1 family contains two members (hnf1a and hnf1b) with only hnf1b being expressed in r5/r6. However, in a microarray screen for hoxb1b-regulated genes (manuscript in preparation), we identified a third hnf1 gene identical to NCBI entry AF250352. While this gene has been previously named hnf1γ, we think it is unlikely to represent a third hnf1 family member since other vertebrates do not appear to have a third hnf1 gene. Instead, this gene may represent a duplicated copy of hnf1b. Indeed, sequence alignment reveals it to be more similar to hnf1b than to hnf1a, and it clusters with hnf1b in a phylogenetic tree (Fig. 2A). Further, this gene is on zebrafish LG21 and hnf1b is on LG15. We find that at least three adjacent genes (aldoc, sfrs1, and dnl2) are duplicated between LG21 and LG15 in zebrafish. In contrast, hnf1b, aldoc, and sfrs1 map to a single chromosome in human (Hs17) and in mouse (Mm11). Lastly, the expression pattern of the novel gene is similar to that of hnf1b (Fig. 2B–F). We conclude that this hnf1 gene likely resulted from the genome duplication reported to have occurred in the teleost lineage²⁰ and have named it hnf1b-like (hnf1bl).

FIG. 2.

A novel hnf1bl gene expressed in r5. (A) Phylogenetic tree demonstrating that hnf1bl clusters with hnf1b genes rather than hnf1a genes. (B–H) Wild-type (B–F), hnf1b^hi2169 (G), and val^b337 (H) embryos were assayed for expression of hnf1bl (B, D), hnf1b (C), hnf1bl + krox20 (E, G, H), or hnf1b + krox20 (F) by in situ hybridization. Note that hnf1b and hnf1bl are detected in blue, while krox20 is detected in red. (B), (C), (G), and (H) are whole mounts while (D), (E), and (F) are flat mounts. Anterior is to the top in all panels. (I–N) Uninjected (I–K) and hnf1bl-injected (L–N) wild-type embryos were assayed for expression of hoxb1a + krox20 (I, L), hoxa3 + krox20 (J, M), or val + krox20 (K, N). Note that hoxb1a, hoxa3, and val are detected in blue, while krox20 is detected in red. All embryos are whole mounts in dorsal view with anterior to the top.

We do not detect hnf1bl expression until 10 hpf, at which point it is expressed in the intermediate mesoderm (Fig. 2B). This is in contrast to hnf1b, which is expressed in both intermediate mesoderm and the caudal hindbrain (including r5/r6) at this stage (Fig. 2C; Sun and Hopkins⁴). Hnf1bl expression appears in r5 by 14 hpf (Fig. 2D, E), a stage when hnf1b expression has already regressed caudally (Fig. 2F), suggesting that hnf1bl might take over the role of hnf1b in r5/r6 after hnf1b is lost in this region. Hnf1bl expression then diminishes by 18 hpf.

Misexpression of hnf1bl by mRNA injection produced a phenotype where hoxb1a expression in r4 is reduced or lost and gene expression from r3 and r5 expands into r4 (Fig. 2L–N). This phenotype is indistinguishable from that observed upon injection of hnf1b mRNA,^2–4,16 demonstrating that hnf1bl has similar activity to hnf1b and raising the possibility that hnf1bl may compensate for the loss of hnf1b activity in hnf1b^hi2169 embryos.

To test whether hnf1bl might act redundantly with hnf1b, we designed antisense morpholino oligonucleotides (MOs) to hnf1bl (see Materials and Methods). Hnf1bl MOs efficiently blocked expression of an Hnf1bl-GFP fusion protein in vivo (Supplemental Fig. 1, available online at www.liebertpub.com/fpd), demonstrating the efficacy of these MOs. We do not observe any effect on r5 gene expression upon injecting hnf1bl MOs into wild-type embryos (at concentrations up to 1 mM; not shown), indicating that hnf1bl is either not required for r5 gene expression or functions redundantly with hnf1b. To test the latter possibility, we examined the effect of injecting hnf1bl MOs into hnf1b^hi2169 embryos. We find that injecting hnf1bl MOs results in 26% (13/50) of hnf1b^hi2169 homozygous embryos lacking r5 gene expression. This is only slightly higher than the 19% (6/32) of uninjected hnf1b^hi2169 homozygous embryos that lack r5 gene expression. Further, coinjecting hnf1bl MOs with hnf1b MOs into wild-type embryos did not produce a more severe effect than injecting hnf1b MOs alone (47% of embryos with little or no r5 staining after coinjection of hnf1bl and hnf1b MOs vs. 57% after injection of hnf1b MOs alone in a representative experiment). Lastly, we examined hnf1bl expression in hnf1b^hi2169 and val^b337 mutant embryos. We find that hnf1bl is lost in both mutants (Fig. 2G, H), demonstrating that hnf1bl expression requires both hnf1b and val. We conclude that hnf1bl acts downstream of hnf1b and val, and that hnf1bl activity cannot account for the residual r5/r6 gene expression in hnf1b^hi2169 embryos.

The hnf1b^hi1843 allele displays a more severe phenotype than the hnf1b^hi2169 allele

Our analysis of the hnf1b^hi2169 allele suggests that either hnf1b is not absolutely required for r5/r6 formation, or residual hnf1b activity persists in hnf1b²¹⁶⁹ embryos. To address this question, we turned to additional hnf1b alleles. Of the three hnf1b alleles identified in the original insertion screen, hnf1b^hi548 was reported as the weakest, while hnf1b^hi2169 and hnf1b^hi1843 gave similar phenotypes that were more severe than the hnf1b^hi548 phenotype.⁴ We therefore focused on the hnf1b^hi1843 allele and find that it has a more severe hindbrain phenotype than hnf1b^hi2169. In particular, hnf1b^hi1843 embryos appear to completely lack krox20 expression in r5 (Fig. 3B–E) and hoxa3 expression in r5/r6 (Fig. 3G–J) compared to hnf1b^hi2169 embryos that showed variable residual expression (Fig. 1). However, we have also noted that some outcrosses of hnf1b^hi1843 fish occasionally reveal detectable r5/r6 gene expression, suggesting that the hnf1b^hi1843 phenotype may be modified on some genetic backgrounds (e.g., Tupfel longfin; not shown). Hnf1b^hi1843 embryos also display robust uniform expression of hoxb1a throughout the caudal hindbrain (Fig. 3L–O), in contrast to hnf1b^hi2169 embryos that sometimes showed reduced expression of hoxb1a in the r5/r6 region (Fig. 1). Our results indicate that hnf1b^hi2169 is hypomorphic to hnf1b^hi1843 and suggest that residual hnf1b activity persists in the hnf1b^hi2169 allele. Notably, the differences between these alleles also rule out the possibility that the residual r5/r6 gene expression in hnf1b^hi2169 embryos is due to maternal hnf1b mRNA.

FIG. 3.

(A–O) Wild-type (A, F, K) and hnf1b¹⁸⁴³ (B–E, G–J, L–O) embryos were assayed for expression of krox20 (A–E), hoxa3 (F–J), and hoxb1a (K–O) by in situ hybridization. All panels are dorsal views of flat-mounted hindbrains with anterior to the top. Rhombomere numbering is indicated in panels (A), (F), and (K). (P) Wild-type (+/+), heterozygous (+/−), and homozygous hnf1b^hi2169 mutant (−/−) embryos were assayed for presence of hnf1b transcripts by RT-PCR. Virus integration site and locations of primers used for RT-PCR analysis are indicated in diagram at top. RT-PCR products were resolved by gel electrophoresis, and the resulting gel is shown at the bottom. (Q) Primer extension analysis of virus-hnf1b transcript. Top panel shows structure of wild-type hnf1b mRNA and indicates retroviral insertion site (triangle) in the hnf1b^hi2169 allele. Bottom panel shows putative viral-mRNA fusion transcript and indicates sequence obtained from primer extension analysis. (R–V) hnf1b^hi2169 embryos injected with an hnf1b splice-blocking morpholino (SPMO) show near-complete loss of krox20 expression in r5 compared to Figure 1B–E.

To explore the basis of the weaker phenotype of the hnf1b^hi2169 allele, we next examined hnf1b transcripts in this mutant. PCR amplification with primers spanning the entire hnf1b open reading frame (ORF) did not detect a full-length hnf1b transcript in homozygous hnf1b^hi2169 embryos (Fig. 3P), in agreement with previous reports.⁴ Primers amplifying the 5′ end of the ORF (upstream of the viral integration site including the dimerization domain; “DD only” in Fig. 3P) also did not yield a product in homozygous hnf1b^hi2169 embryos, suggesting that transcripts originating at the endogenous hnf1b promoter may not be stable in hnf1b^hi2169 mutants. In contrast, a series of primer pairs amplifying regions of the hnf1b transcript downstream of the integration site yielded fragments encoding the POU domain, the homeodomain, and the activation domain (“POU to end” and “AtH to end” in Fig. 3P). Hence, a partial hnf1b transcript is present in hnf1b^hi2169 embryos. Using primer-extension analysis we found that this partial transcript extends at least 43 bp into the viral long terminal repeat (LTR) sequence (Fig. 3Q). Since retroviral LTRs contain promoter activity, it is possible that the transcript originates in the 3′ LTR of the integrated retrovirus. Although we cannot rule out the possibility that the transcript originates at the endogenous hnf1b promoter and uses a splice site within the virus to create a fusion transcript, this appears unlikely since the transcript appears to lack the 5′ end of hnf1b (Fig. 3P). Since we lack antibodies to zebrafish Hnf1b, we cannot test directly whether this fusion transcript produces a protein. Instead, we generated a splice MO (sMO) that targets the intron 1/exon 2 junction, which is located downstream of the virus integration site in exon 1. The intron 1/exon 2 boundary corresponds to the N-terminal end of the POU domain, and the sMO is therefore expected to interfere with splicing of the fusion transcript and to disrupt translation of any protein produced from the fusion transcript. Accordingly, the sMO reduces the level of the properly spliced transcript by approximately six- to sevenfold (Supplemental Fig. 2, available online at www.liebertpub.com/fpd). We find that the hnf1b sMO reduces residual krox20 expression in r5 of hnf1b^hi2169 embryos (Fig. 3R–V, compare to Fig. 1B–E). We conclude that a fusion transcript in hnf1b^hi2169 embryos is responsible for residual gene expression in r5.

Hoxb1b induces ectopic r5-r6 gene expression in hnf1b^hi2169 and hnf1b^hi1843, but not in val^b337 embryos

We have reported that hoxb1b (acting together with its meis and pbx cofactors) is required for r5/r6 gene expression^16,21,22 and can induce ectopic expression of r5/r6 genes in zebrafish embryos.²³ To determine if hnf1b is absolutely required for induction of r5/r6 gene expression, we tested whether hoxb1b can induce r5/r6 gene expression in hnf1b^hi2169 or hnf1b^hi1843 embryos. As expected, coinjection of hoxb1b and meis3 mRNA (pbx mRNA need not be injected as pbx2 and pbx4 are ubiquitously expressed up to 24 hpf^24–26) induces ectopic expression of r5/r6 genes anterior to the hindbrain in wild-type embryos (Fig. 4A, E). Hoxb1b/meis3 injection also induces ectopic hoxa3 expression in homozygous hnf1b^hi2169 (19/19) and homozygous hnf1b^hi1843 (8/9) mutant embryos (Fig. 4B, C), but not in homozygous val^b337 mutant embryos (0/7; Fig. 4D). Similarly, hoxb1b/meis3 induces ectopic expression of both val (Fig. 4E–G) and krox20 (not shown) in both hnf1b^hi2169 (13/13 for val and 10/10 for krox20) and hnf1b^hi1843 (11/12 for val) mutant embryos. Notably, while hoxb1b/meis3 can induce ectopic r5/r6 gene expression, it cannot restore normal gene expression to r5/r6 of hnf1b mutant embryos. This result indicates that hnf1b is required for induction of r5/r6 gene expression in the hindbrain, but not further anteriorly in the embryo. In contrast, val is required for induction of r5/r6 gene expression throughout the embryo.

FIG. 4.

hoxb1b activates r5/r6 gene expression independently of hnf1b in the anterior embryo, but not in the hindbrain. Wild-type (A, E), hnf1b^hi2169 (B, F), hnf1b^hi1843 (C, G), and val^b337 (D) embryos were injected with hoxb1b and meis3 mRNA and assayed for expression of hoxa3 (A–D) and val (E–G). Ectopic gene expression anterior to the hindbrain is indicated by asterisks. All panels are flat-mounted dorsal views of the anterior embryo with anterior to the top.

Discussion

Previous work led to the suggestion that hnf1b is required for activation of r5/r6 gene expression in the zebrafish hindbrain. However, the hnf1b^hi2169 retroviral insertion allele used in most studies shows residual r5/r6 gene expression (Fig. 1), indicating that this may not represent a null allele, or alternatively, that hnf1b is not absolutely required for r5/r6 gene expression. We detect a virus-hnf1b fusion transcript in hnf1b^hi2169 embryos and demonstrate that MO-mediated knockdown of this transcript produces a more severe phenotype with almost complete loss of r5/r6 gene expression (Fig. 3). While we have not detected a protein produced from this transcript, we note that translation would likely be initiated at one of two methionines (positions 120 and 131) at the N-terminal end of the POU domain. In addition, a second retroviral insertion allele (hnf1b^hi1843) also displays almost complete loss of r5/r6 gene expression (Fig. 3). We conclude that hnf1b²¹⁶⁹ is not a null allele and that residual r5/r6 gene expression in hnf1b^hi2169 embryos is likely mediated by a virus-hnf1b fusion transcript. Hence, our results are consistent with hnf1b being required for r5/r6 gene expression. Surprisingly, we detect a virus-hnf1b fusion transcript also in hnf1b¹⁸⁴³ embryos (not shown), but this transcript may not be fully active since the retrovirus integrated into exon 2 (which encodes the POU homeodomain) in the hnf1b^hi1843 allele.

Using hoxb1b-mediated induction of r5/r6 gene expression as an assay, we also demonstrate that hnf1b is not required for ectopic expression of r5/r6 genes in the anterior embryo (Fig. 4). In contrast, hnf1b appears to be required for restoration r5/r6 gene expression in the hindbrain (Fig. 4). This result indicates that hnf1b plays a unique role in the hindbrain. While it is not completely clear what this role might be, it has been demonstrated that hnf1b is required to repress hoxb1a expression in the caudal hindbrain.^2–4,16 Since hoxb1a is not expressed in the anterior central nervous system (CNS), such a repressive hnf1b activity might not be needed outside the hindbrain. An interesting corollary to this hypothesis is that the primary role for hnf1b in r5/r6 may not be to activate r5/r6 genes, but to repress expression of genes such as hoxb1a that promote competing rhombomere fates. Indeed, there are reports of hnf1b proteins acting as repressors.²⁷

Val is thought to act downstream of hnf1b in the induction of r5/r6 gene expression, and we therefore used the same assay to examine the requirement for val in hoxb1b-mediated induction of r5/r6 gene expression. In contrast to the situation for hnf1b, val function is absolutely required for induction of ectopic r5/r6 gene expression (Fig. 4). This finding is consistent with a role for val in activation of r5/r6 gene expression and underscores the fact that vhnf1 and val may act by distinct mechanisms to promote r5/r6 gene expression.

Materials and Methods

Zebrafish lines

Hnf1b^hi2169 and hnf1b^hi1843 fish were obtained from N. Hopkins and the Zebrafish Resource Center. Val^b337 fish were obtained from C. Moens.

Plasmid morpholinos and oligonucleotide primers

A full-length hnf1bl cDNA clone was purchased from Open Biosystems (Huntsville, AL). For mRNA synthesis, cDNAs were cloned into the pCS2MT vector and mRNA prepared using the mMessage mMachine kit (Ambion, Austin, TX). Translation start site MOs to hnf1b (5′-CTAGAGAGGGAAATGCGGTATTGTG-3′) and hnf1bl (5′-CTTGGACACCATGTCAGTAAA-3′) as well as an sMO to hnf1b (5′-TCCTCCCTGAAAAGATCGGAAACAT-3′) were purchased from Genetools (Philomath, OR)/Open Biosystems. PCR primers for the amplification of hnf1b cDNA sequences were as follows: Full length forward 5′-CACAATACCGCATTTCCCTCTCTAG-3′; Full length reverse 5′-GTCACTAAATTGGGCGCCATGTTGATCA-3′; DD reverse 5′-CTTCATCTCCGGATAACTTCCCTTTAC-3′; POU forward 5′-CATGATCAAAGGCTACATGCAGCAGCAC-3′; AtH forward 5′-GCGCCACCATGTTAGACAAAGGAAATCAG-3′. The 5′ RACE was carried out using the SMART RACE kit (Clonetech, Mountain View, CA), primed with oligo 5′-CATCATCACCTGGCTGGACCATACCTT-3′, and amplified with nested primer 5′-GATCTCCCGCTGTTTCCTCACATACCA-3′. Hnf1b^hi2169 and hnf1b^hi1843 were genotyped by detection of the retroviral insertion in the hnf1b gene using primers 5′-CACAATACCGCATTTCCCTCTCTAG-3′, 5′-TCCGGATAACTTCCCTTTACTGTG-3′, and 5′-CTGTTCCATCTGTTCCTGAC-3′ for hnf1b^hi2169 and primers 5′-TTCCTATGTAATTGTGTCCGATGATAG-3′, 5′-CAAGCAGGCTCAATGGCAGC-3′, and 5′-GCTAGCTTGCCAAACCTACAGGT-3′ for hnf1b^hi1843. Val^b337 fish were genotyped using primers 5′-CCCGCAGACGTTAAGCCTCAC-3′ and 5′-GATCGCGCCGTACTGGTGTT-3′ for PCR amplification followed by digestion with Pvu II. The efficacy of the hnf1b sMO was tested with primers PCR1 5′-ACCGCCAATTCTCAAAGAGCTC-3′, PCR2 5′-GCGACAGGTGCGACTGATTCAG-3′, and PCR3 5′-CAATTGACCGCACTTGCAAAAT-3′.

Microinjections and in situ hybridizations

mRNA injections were carried out as reported¹⁶ using 140 ng/μL hnf1bl, 166 ng/μL hoxb1b, 166 ng/μL meis3, 0.1–1 mM hnf1bl translation MO, 1 mM hnf1b translation MO, and 2 mM hnf1b sMO. In situ hybridizations were carried out as reported previously.²²

Acknowledgments

We gratefully acknowledge Nancy Hopkins and Cecilia Moens for providing zebrafish lines and Letitiah Etheridge for expert assistance with in situ hybridizations and genotyping.

Disclosure Statement

No competing financial interests exist.